Digital History

Relevance of Machine Learning and Artificial Intelligence to History


First published on August 24, 2019. Last updated on February 20, 2021.

Learning Objectives

  • Students will be introduced to concepts and advances in machine learning and artificial intelligence.
  • The use of AI as an aide to historians and an image recognition and classification tool will be discussed.

What Is Machine Learning and Artificial Intelligence?

Machine learning is essentially telling a computer program that particular items of data are related to each other. For example, you could tell the program the width of a petal and the corresponding name of a plant variety. The program will then be able to make predictions of variety if you provide it petal widths in the future.

Machine learning is a form of artificial intelligence (AI). Artificial intelligence come in several forms, but more sophisticated machine learning is a common form. For example, if you show the program a series of photos, and then provide a tag for each photo (such as flower, dog, house), eventually the program will be able to recognize future photos.

Some experts say that AI will someday be able to completely replace thinking by humans. Other experts disagree, and say that while AI is good at some tasks, that it cannot reproduce all of human thinking processes. So far, the latter are correct, but as AI become further developed, only time will tell who is ultimately correct.

For historians, AI can be useful in going though large batches of text documents and photos and recognizing items of interest. Be warned that, like human minds, AI programs are not perfect. They make mistakes.

How AI and Machine Learning Work

Many AI and machine learning work by comparing something new to something known (or a set of known things). If the new item is sufficiently similar to a known item, then the program considers the new item to be of the same sort as the old item. For example, consider a flower classification system, that figures out whether a new flower is a daisy or rose. If the new flower has radially-extending petals, the program would probably consider it to be a daisy.

However, most programs don’t work on such a clear-cut, rules basis. Rather they work on a statistical basis, and must be trained. For example, a human will load photos of daisies and roses into the system, and indicate the flower type. The system will learn from this training. This way, cases that deviate from the ideal case can still be identified, for instance, a daisy viewed from on edge might still be identified successfully.


Natural Intelligence As An Anology

Humans and other animals have natural intelligence that is capable of learning. (Plants can learn, but the mechanism is much different than for animals and computers). For example, animals learn what to eat or not eat based upon taste corresponding reactions to food. Animals and people learn to recognize images. While some types of images seem to be hardwired into the brain (e.g. how cats react to snake-like objects such as strings and cucumbers). Others are learned, such as letters and words for people. Think about how you learn to recognize and react to images. Think about how movies train people, even within just a few hours.

Resources-Iris Data Sets


  • Tensorflow open source learning platform
  • University of Wiakato Weka is a machine learning tool
  • IBM Watson expert system platform for research
  • PyTorch AI library for Python
  • Lobe AI platform


  1. Teams of students will debate whether AI or “history-bots” will replace historians.
  2. Try out Weka (for students with time and patience. The download and set-up process may take awhile, and some of the concepts may be odd at first, but this is one of the simpler machine learning/AI platforms).


Content is copyright the author. Layout is copyright Mark Ciotola. See for further notices.